Yiuno

Why This System Exists

5 min read

Why This System Exists

A knowledge system for learning by building with AI. This page explains the problem it solves, how it’s structured, and how to navigate it.

The problem

You ask AI to build something. It works. Then something breaks and you can’t fix it — because you have no mental model of what was built.

graph LR
    A[You have an idea] -->|ask AI| B[AI builds it]
    B -->|looks right| C[It works!]
    C -->|something breaks| D[You can't fix it]
    D -->|ask AI again| E[AI rewrites everything]
    E -->|different bugs| D

    style D fill:#d9534f,color:#fff
    style E fill:#d9534f,color:#fff

This is the one-shot problem. AI produces impressive first drafts, but iteration requires understanding.

The intent paradox

Good AI output requires good instructions. Good instructions require understanding. But you’re using AI because you don’t understand yet.

How do you bootstrap understanding for a domain you haven’t encountered?

That’s the question this system answers.

The insight

AI is the most powerful learning tool ever created — but only when combined with structure that preserves effort where it matters.1 Without structure, AI creates an illusion of competence.2 With structure, it accelerates genuine mastery.

This system provides that structure. It gives you just enough understanding at each step to direct the next — bootstrapping your intent incrementally until you don’t need the scaffolding anymore.3

How it works

The system is a graph of interconnected knowledge — not a textbook with a fixed reading order.

graph TD
    SD[Software Development] --> SA[Architecture]
    SD --> CSM[Client-Server Model]
    SD --> ID[Iterative Development]
    CSM --> FE[Frontend]
    CSM --> BE[Backend]
    CSM --> API[APIs]
    CSM --> DB[Databases]
    SA --> SoC[Separation of Concerns]
    SA --> CI[Contracts]
    API --> EP[Endpoints]
    API --> HM[HTTP Methods]

    style SD fill:#4a9ede,color:#fff

It’s made of three things:

1. Concept cards

Self-contained articles that each teach one concept from zero. Every card has a definition, analogies, a Mermaid diagram, a comprehension gate, and links to related cards.

Think of it like a museum

Each card is a display case. It has a label, a description, a model, and arrows pointing to related exhibits. You can wander in any order — but each exhibit is complete on its own.

See the APIs card for a live example of what a concept card looks like. See card-template for the full structure every card follows.

2. Learning paths

Narrative articles that guide you through a curated sequence of concept cards. You can read a path start to finish and learn something — even without opening a single card. The cards are depth-on-demand.

graph LR
    E[Entry] --> P1[Part 1] --> P2[Part 2] --> P3[Part 3]
    P3 --> GATE[Gate]
    GATE -->|path A| N1[Next path]
    GATE -->|path B| N2[Different path]

    style E fill:#4a9ede,color:#fff
    style GATE fill:#e8b84b,color:#fff

Every path starts from a reader profile and ends with multiple exits. See from-zero-to-building for a live example.

3. Comprehension gates

The key innovation. At the end of every card and every path, there are questions that test whether you’ve built a real mental model — not just read the words.

Gates exist because AI makes it too easy to skip understanding.1 They’re a commitment to yourself: stop, test, then proceed.

The principle

Comprehension precedes progression. If you can’t explain it, you don’t understand it yet.

The knowledge graph

Cards connect through four relationship types:

RelationshipWhat it meansExample
ParentBroader concept aboveAPIs → Client-Server Model
ChildrenGranular concepts belowAPIs → Endpoints, HTTP Methods
PrerequisitesUnderstand these firstAPIs → Contracts and Interfaces
RelatedLateral connectionsAPIs ↔ Frontend, Backend

Cards are classified by scope:

LevelScopeExample
DomainBroadest fieldSoftware Development
DisciplineA specialisationWeb Development
TopicA course-sized areaAPIs
ConceptA chapter-sized ideaHTTP Methods
AtomicOne key distinctionGET vs POST

For the full taxonomy rules, see taxonomy.

The dynamic load shift

As you progress, the balance between you and AI shifts:

graph LR
    E[Early] --> M[Mid] --> L[Late]

    style E fill:#d9534f,color:#fff
    style M fill:#e8b84b,color:#fff
    style L fill:#5cb85c,color:#fff
PhaseYouAI
EarlyAbsorb, ask questionsExplains, generates examples
MidDirect, make decisionsAssists, follows instructions
LateOrchestrate, architectExecutes your design

The system makes this visible. Each gate you pass is proof that the balance has shifted — that more of the understanding lives in you, not in the AI.3

How to navigate

Browse the graph

Open Obsidian’s graph view. Click any node. Follow links.

Follow a path

Pick a learning path that matches your level. Read it. Do the gate. Choose your next path.

Explore from a concept

Land on any concept card. Follow “Concepts to explore next” to go deeper or wider.

Add knowledge

Use card-template to create a card for something new you encounter. Connect it to existing cards. The graph grows with you.

Check your understanding

Where to start

I've never built software before

Start with from-zero-to-building. It teaches how software is structured using everyday analogies — no code, no jargon.

I know the basics and want to build something

Browse the concept cards and follow what interests you, or try agentic-design for an intermediate path.

I want to understand how this system is built

Read the system docs: taxonomy · card-template · path-template · stub-index

Sources

Further reading

Resources

Footnotes

  1. Bjork, E. L. & Bjork, R. A. (2011). Making Things Hard on Yourself, but in a Good Way. UCLA. ↩ ↩2

  2. Nazri, N. & Abdul Rani, N. (2026). The Illusion of Competence: AI-Assisted Learning. IJRSI. ↩

  3. Wood, D., Bruner, J. S. & Ross, G. (1976). The Role of Tutoring in Problem Solving. Journal of Child Psychology and Psychiatry. ↩ ↩2

Graph View

Backlinks